claim
active
claim:width-buys-features-in-order-of-utility-so-only-larger-models-learn-infrequent-or-complex-tasksWidth buys features in order of utility, so only larger models learn infrequent or complex tasks
Interpretation of Theorem 3.
Source paper
extracted_from(2026) · Jing Huang · Daniel Wurgaft · Rachit Bansal · Laura Ruis +5
Neighborhood — ranked by edge-count
Papers (1)
paper
Findings (4)
finding
- Rank-1 tasks (K=32, β=2) show empirical transition width matching N_crit(k)=k staircase almost perfectlyassociated_withsupportsSharp test of utility ordering.
- Fig. 2 verification.
- Fig. 5.
- Fig. 6a.
Claims (1)
claim
- Core phenomenological argument.
Related by similarity (8)
cosine ≥ 0.65 · no typed edgeEntities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.
- Width-N minimizer spans top-N eigenspace of M.
- Core framing.
- As we make a model wider, do we just get more features, or the same features packed less densely?question0.790Fresh-traction question.
- Authors argue features are model properties because logit effects and ablations are consistent with feature interpretations
- Hypothesis box in §3.2.
- Inverse scaling interpretation.
- Can we transform a model's features into an inexpensive, dense signal for learning open-ended tasks?question0.784Motivating question.
- Mechanism from Theorem 4.